Re: [PATCH 3/4] xen: set the right flags when enabling interrupts for 8250

[Julien Grall <julien.grall@linaro.org>]

On 07/26/2013 12:01 PM, Chen Baozi wrote:
> 
> On Jul 26, 2013, at 6:42 PM, Julien Grall <julien.grall@linaro.org> wrote:
> 
>> On 07/26/2013 04:31 AM, Chen Baozi wrote:
>>>
>>> On Jul 26, 2013, at 9:14 AM, Chen Baozi <baozich@gmail.com> wrote:
>>>
>>>>
>>>> On Jul 25, 2013, at 8:33 PM, Julien Grall <julien.grall@linaro.org> wrote:
>>>>
>>>>> On 07/25/2013 12:31 PM, Chen Baozi wrote:
>>>>>>
>>>>>> On Jul 25, 2013, at 7:17 PM, Julien Grall <julien.grall@linaro.org> wrote:
>>>>>>
>>>>>>> On 07/25/2013 10:14 AM, Chen Baozi wrote:
>>>>>>>
>>>>>>>> Besides this test, I noticed that we set a timer to poll UART after enabling
>>>>>>>> the interrupt. And the timer initialization codes are platform specific, which
>>>>>>>> should be implemented in xen/arch/arm/platforms/omap5.c for example. However,
>>>>>>>> I haven't done this yet. (I used to plan to turn to this work after UART porting
>>>>>>>> has been done.) Is it a possible factor that may cause my issues?
>>>>>>>
>>>>>>> Xen uses the arch timers to handle the time. On some platform (for
>>>>>>> instance the Arndale), it's not enabled by default.
>>>>>>>
>>>>>>> If the omap5 boot with the arch timers enabled, you don't need to
>>>>>>> implement the init_time callback.
>>>>>>>
>>>>>>> Do you know if Xen receives timer interrupts?
>>>>>>
>>>>>> Any way that I could confirm this?
>>>>>>
>>>>>> There is a line of boot message:
>>>>>>
>>>>>> "Using generic timer at 0 KHz."
>>>>>>
>>>>>
>>>>> It seems the arch timer is not configured/enabled. Do you know if U-boot
>>>>> enables it?
>>>>
>>>> I could see clocks.c/clocks.o in its u-boot sources, but I don't think U-boot has already enabled it, for I could read such lines from Linux kernel boot messages:
>>
>> Are you trying to boot Linux on bare metal or as dom0?
> 
> bare metal.
> 
>>
>>>> """
>>>> OMAP clockevent source: GPTIMER1 at 32768 Hz
>>>> sched_clock: 32 bits at 32kHz, resolution 30517ns, wraps every 131071999s
>>>> OMAP clocksource: 32k_counter at 32768 Hz
>>>> arch_timer: No interrupt available, giving up
>>
>> Do you have a "arm,armv7-timer" compatible node in your DTS with
>> interrupts? Here Linux can't find an appropriate interrupt.
> 
> You reminds me that the DTS shipped along with the board has issues on "arm,armv7-timer". I have corrected it for xen but leave the bare metal booting dtb the original status.
> 
>>
>>>> omap5_realtime_timer_init: arch_timer_register failed -22
>>>>
>>>>>
>>>>> It's not clear to me, how you need to implement init_time. Do you have a
>>>>> datasheet with the board? If yes, is there a section for the section timer?
>>>>
>>>> Yes, TI has published an 49.6MB Technical Reference Manual online with 88 pages section about timer.
>>>
>>> Wait. I think I might make a mistake. The "arch timer" is different from "timer" as devices, right? The timer section in Technical Reference Manual seems to be the latter one, for it is  connect to the Level-4 interconnect of OMAP5432 and described as "ti,omap5430-timer" in DTS. I think the "arch timer" you mentioned should refer to "arm,armv7-timer", right?
>>
>> By "arch timer" I mean "arm,armv7-timer".
> 
> Thanks. So I think the codes to start with it is linux/drivers/clocksource/arm_arch_timer.c: arch_timer_init(). The linux/arch/arm/mach-omap2/timer.c is for the timer device "ti,omap543-timer".

arch_timer_init is already implemented in init_xen_time
(xen/arch/arm/time.c).
On some board, the arch timer is not enabled by default and need a
"workaround" for this purpose.

For instance on the Arndale, we need to enable the MCTimer. I think you
need to do something like that for the OMAP.

Do you know if the boot CPU already start in HYP mode? If no, did you
set up correctly the timer in mode_switch.S?

-- 
Julien